Hospital Out Patient Visit Forecasting using Gated Recurrent Unit

Recently, people have become more interested in their healthcare and well-being, resulting in a significant increase in patient flow for healthcare providers. To provide quality care, hospitals must manage patient flow and resources effectively. Hospitals need to forecast daily patient flow in the Out Patient department to ensure they have enough resources to provide care to all patients, manage resources efficiently, reduce wait times, and improve patient satisfaction. This study is about optimizing patient visit forecasting using deep learning techniques on univariate time series data. The LSTM-based deep learning approach is a reliable module for forecasting such data, however, an enhanced version of LSTM called GRU produced even more effective results. The experimental data from a private hospital’s daily outpatient visits over the last 5 years verified that the GRU model achieved a higher accuracy of 92.42% before 1 minute 29 seconds compared to the LSTM model whose accuracy of 87.15% on an average time of 13minute 8 seconds.